Coffee beans are roasted to bring out their unique aroma and taste. When a user tries to roast coffee beans by a roasting machine, he/she can use a roasting curve in which temperature is set to the vertical axis and time is set to the horizontal axis. In the roasting of coffee beans, the user needs to select an appropriate roasting curve depending on the desired characteristics. However, it is difficult for him/her to know what characteristics he/she wants to produce and what roasting curve he/she should use to roast the coffee beans to produce the characteristics. Therefore, roasting is a skilful technique performed by experts. In this study, we collected various data related to roasting. Furthermore, we developed a learning model that outputs color information of the resulting beans from the input information on roasting using multiple machine learning algorithms. Finally, we compared and verified the accuracy of the model.

In order to estimate recent surface displacement in Bangkok, we applied persistent SAR interferometry using ALOS/PALSAR images acquired October 2007 to December 2010. Previously, land subsidence due to excessive ground water pumping has been reported in this region. However, we detected that ground surface has rebounded during 3 years of the observation period. This rebounded surface can be interpreted as the response on the ground water recovery to confined aquifer due to the decrease in ground water usage. Since the rebounded areas correspond to past ground subsidence areas inferred from differential SAR interferometry using JERS-1/SAR images, it can be inferred that ground water recovery has especially occurred at the preceding subsidence areas.

It is common for clients using Twitter to adopt displays that indicate their posts (tweets) as textboxes, and they are individually and perpendicularly lined up in order. When several people discuss a particular topic, they frequently post tweets as "replies". However, it is difficult for users in these types of displays to obtain the information that is required and beneficial to understand to make progress in particular discussions. Consequently, we propose a system, which we evaluated, of visualizing the relations between tweets by using digraphs.

This paper proposes a 3D graphics simulation environment for the analysis and design evaluation of a layout reconfiguration process of the reconfigurable manufacturing system (RMS). Manufacturing cells (MCs) in the RMS are positioned to minimize a total distance of transport paths for a given manufacturing layout. After the end of current manufacturing process, the MCs start to move to each position in the optimal layout of next manufacturing process. The positioning and timing of the MCs in the reconfiguration process were designed by a reconfigurable process design method based on particle swarm optimization (PSO). From the simulation results, the effectiveness of the reconfiguration process design method based on PSO was confirmed

People recognize the quality of a product while in operation by hands or fingers. The operation feeling by hands or fingers is one of the important indexes for the high-grade products. However, skilled inspectors are used to inspect some products because automatic inspection is technologically difficult or too high in cost. This paper looks at a system for inspection of the quality of a product’s reaction force characteristics. This system, until now considered difficult to realize, automates the inspection method utilizing the touching of an inspector's finger. Neural network classifier is applied to the system for products to learn an inspector's finger judgment. We provide an input layer of a neural network classifier with nodes corresponding to time-and frequency-domain features of reaction forces of a product and an output layer with three nodes corresponding to a judgment; being one of non-defective, defective, or unable to judge. From experimental results, the effectiveness of the proposed neural network classifier has been clarified.

A new reconfigurable parts feeding system using arrayed vibratory units made by stereolithography has been developed and investigated. The parts feeding motion is achieved by vibration of an inclined pin grid array using piezoelectric actuators. Experiments on moving objects have been successfully performed. The vibratory unit has a large number of inclined pins, each a height of 3 mm and diameter of 0.5 mm. The motions of the parts feeding system has been demonstrated experimentally, and lifetime measurements demonstrated the long-term stability of the system. The functionality of the photopolymer resin pins is unaffected through 110 days continuous use.

This paper proposes a 3D graphics simulation environment for the analysis and evaluation of reconfigurable manufacturing systems. The reconfigurable manufacturing systems consist of transport robots; input stations, output stations, movable manufacturing cells, and objects to process. The layout of the manufacturing cells and the allocation of transport robots are optimized using particle swarm optimization. The effectiveness of the simulator is validated through simulation results of several manufacturing cases.

Looks at a system for inspection of the quality of a probe's reaction force characteristics. This system, until now considered difficult to realize, automates the inspection method utilizing the touching of an inspector's finger. Wavelet transformation and a neural network (NN) approach are applied to the system for probes to learn an inspector's finger judgment. We provide an input layer of a NN with thirty-three nodes corresponding to a time series of reaction forces of a probe and an output layer with one node corresponding to a judgment; being one of non-defective, defective, or unable to judge. From experimental results, the effectiveness of the system has been clarified.